Substrate flow in catalases deduced from the crystal structures of ...

Substrate Flow in Catalases Deduced from the CrystalStructures of Active Site Variants of HPII fromEscherichia coliWilliam Melik-Adamyan,1 Jeronimo Bravo,2 Xavier Carpena,2 Jack Switala,3 Maria J. Mate,2 Ignacio Fita,2

and Peter C. Loewen3*1Institute of Crystallography, Russian Academy of Sciences, Moscow, Russia2Instituto Biologıa Molecular Barcelona (C.S.I.C.), Barcelona, Spain3Department of Microbiology, University of Manitoba, Winnipeg, Manitoba, Canada

ABSTRACT The active site of heme catalases isburied deep inside a structurally highly conservedhomotetramer. Channels leading to the active sitehave been identified as potential routes for sub-strate flow and product release, although evidencein support of this model is limited. To investigatefurther the role of protein structure and molecularchannels in catalysis, the crystal structures of fouractive site variants of catalase HPII from Esche-richia coli (His128Ala, His128Asn, Asn201Ala, andAsn201His) have been determined at ;2.0-Å resolu-tion. The solvent organization shows major rear-rangements with respect to native HPII, not only inthe vicinity of the replaced residues but also in themain molecular channel leading to the heme distalpocket. In the two inactive His128 variants, continu-ous chains of hydrogen bonded water moleculesextend from the molecular surface to the hemedistal pocket filling the main channel. The differ-ences in continuity of solvent molecules betweenthe native and variant structures illustrate howsensitive the solvent matrix is to subtle changes instructure. It is hypothesized that the slightly largerH2O2 passing through the channel of the nativeenzyme will promote the formation of a continuouschain of solvent and peroxide. The structure of theHis128Asn variant complexed with hydrogen perox-ide has also been determined at 2.3-Å resolution,revealing the existence of hydrogen peroxide bind-ing sites both in the heme distal pocket and in themain channel. Unexpectedly, the largest changes inprotein structure resulting from peroxide bindingare clustered on the heme proximal side and mainlyinvolve residues in only two subunits, leading to adeparture from the 222-point group symmetry of thenative enzyme. An active role for channels in theselective flow of substrates through the catalasemolecule is proposed as an integral feature of thecatalytic mechanism. The Asn201His variant of HPIIwas found to contain unoxidized heme b in combina-tion with the proximal side His–Tyr bond suggestingthat the mechanistic pathways of the two reactionscan be uncoupled. Proteins 2001;44:270–281.© 2001 Wiley-Liss, Inc.

Key words: crystal structure; catalase; catalysis;protein channels

INTRODUCTION

Catalase (hydrogen peroxide:hydrogen peroxide oxi-doreductase, EC 1.11.1.6), present in most aerobic organ-isms, is a protective enzyme that removes hydrogen perox-ide before it can decompose into highly reactive hydroxylradicals. The characteristic catalase activity—the cata-lytic reaction—uses hydrogen peroxide as both an electrondonor and an electron acceptor, as summarized in theoverall reaction 1:

2H2O23 2H2O 1 O2 (1)

This reaction takes place in two distinct stages: in thefirst stage (reaction 2) the resting state enzyme is oxidizedby hydrogen peroxide to an intermediate, compound I,which in the second stage (reaction 3) is reduced back tothe resting state by a second hydrogen peroxide:

Enz ~Por–FeIII! 1 H2O23 Cpd I ~Por1•–FeIVAO! 1 H2O(2)

Cpd I ~Por1•–FeIVAO! 1 H2O23

Enz ~Por–FeIII! 1 H2O 1 O2 (3)

The structures of heme catalase isolated from sevendifferent sources have now been solved, including thosefrom bovine liver,1,2 human erythrocytes,3,4 Penicilliumvitale,5,6 Saccharomyces cerevisiae,7 Proteus mirabilis,8 Mi-crococcus lysodeikticus,9 and Escherichia coli,10,11revealing a

Grant sponsor: DGICYT (Spain); Grant number: BIO099-0865;Grant sponsor: European Union; Grant number: CHGE-CT93-0040;Grant sponsor: Natural Sciences and Engineering Research Council ofCanada (NSERC); Grant number: RGP9600; Grant sponsor: NATO;Grant number: SA-5-2-05; Grant sponsor: MEC (Spain); Grant num-ber: SAP 1998-0087.

William Melik-Adamyan and Jeronimo Bravo contributed equally tothis work.

Jeronimo Bravo is currently at the MRC Laboratory of MolecularBiology, Hills Road, Cambridge CB2 2QH, UK.

*Correspondence to: Peter Loewen, Department of Microbiology,University of Manitoba, Winnipeg, MB R3T 2N2 Canada. E-mail:[email protected]

Received 8 December 2000; Accepted 22 March 2001

Published online 00 Month 2001

PROTEINS: Structure, Function, and Genetics 44:270–281 (2001)

© 2001 WILEY-LISS, INC.

common, highly conserved, core structure in all enzymes.The heme active center in catalase HPII from E. coli shareswith all other catalases the structural features that areimportant for catalysis, in particular the presence andrelative orientation of a histidine (His128) and an aspara-gine (Asn201) on the distal side of the heme and a tyrosine(Tyr415) on the proximal side (Fig. 1A). However, HPII isunique among catalases in having two posttranslationalmodifications in the vicinity of the active center, which areself-catalyzed after the adoption of the final quaternarystructure. One of the modifications, also found in someother large subunit catalases,12 is an oxidized heme group,in the form of a cis-spirolactone, termed heme d. Thesecond modification, so far unique to HPII, is a covalentbond between the Nd of His392 and the Cb of Tyr415, theproximal side fifth ligand of the heme.13 Both modifica-tions seem to require some degree of catalytic activity, andit has been proposed, based on the properties of a numberof HPII variants11, that compound I acts as an initiator ofthese modifications with a possible mechanistic relation-ship between the two reactions.

A number of catalase HPII variants have been con-structed that incorporate changes into the catalytic resi-dues, His128 and Asn201, on the distal side of the heme.Replacement of His128 with any one of a number ofresidues results in variants with no detectable activity orin variants that are defective in folding, such that noprotein accumulates.14 Replacement of Asn201 with aspar-tate, glutamine, alanine, or histidine causes a reduction inspecific activity to less than 10% of wild-type levels, but inno case does activity become nondetectable.14 On the basisof these data, it was concluded that His128 is essential forthe enzymatic activity, while Asn201 facilitates catalysis,but is not essential.

Despite all the structural and biochemical informationabout catalases that has accumulated over 100 years, anumber of basic questions remain unanswered, amongthem: (1) how efficient substrate flow toward the deeplyburied active centers is achieved; and (2) the basis of thespecificity for hydrogen peroxide. To address these issues,a structural determination of variants of the active siteresidues His128 and Asn201 was undertaken. The com-plete absence of activity in the His128 variants alsoallowed the isolation and structural analysis of complexesbetween the protein and hydrogen peroxide. Data obtainedprovide significant insights into the roles of molecularchannels and protein structure in the mechanism of cataly-sis.

METHODS

The four HPII variant proteins analyzed in this workwere purified from the catalase deficient E. coli UM255 aspreviously described.14 Crystals were obtained at roomtemperature by the vapor diffusion hanging drop methodat a protein concentration of ;15 mg/ml over a reservoircontaining 15–17% PEG 3350, 1.6–1.5 M LiCl and 0.1 MTris-HCl, pH 9 as previously described for native HPII.11

Crystals were monoclinic space group P21 with one tet-rameric molecule in the crystal asymmetric unit. Diffrac-

tion data for the His128Ala, His128Asn, and Asn201Alavariants were obtained from crystals transferred to asolution containing 30% PEG 3350 and flash-cooled with anitrogen cryo-stream, yielding the following unit cell pa-rameters: a 5 93.0 Å, b 5 132.3 Å, c 5 121.2 Å and b 5109.3°. Data collection for the isomorphic crystals of theAsn201His variant was carried out at room temperature,yielding the following unit cell parameters: a 5 95.2 Å, b 5134.7 Å, c 5 124.4 Å and b 5 109.4°. The four diffractiondata sets were processed using the program DENZO andscaled with program SCALEPACK.15 In this study, 5% ofthe measured reflections in every data set was reserved forRfree monitoring during automatic refinement (Table I).

Refinement of the four variants was carried out with theprogram XPLOR using standard protocols.16 For eachvariant, the starting model was the HPII structure refinedto 1.9 Å.13 Coordinates were readjusted to the correspond-ing unit cells using the option balloon of XPLOR, and theposition of the molecule was then refined with some cyclesof rigid body refinement. Altered residues and neighborsolvent molecules were always omitted in the startingmodels. Refinements were completed using the programREFMAC17 with solvent molecules modeled either auto-matically using ARPP18 or manually with the graphicprogram O.19 Solvent molecules were only introducedwhen they corresponded to the strongest peaks in thedifference Fourier maps that could make at least onehydrogen bond with atoms already in the model. In thefinal rounds of refinement, the four subunits were treatedindependently with the bulk solvent correction appliedand the whole resolution range available used for eachvariant. To ensure the consistency of the differences foundin the HPII variant structures, the native HPII was alsore-refined using the same procedures (Table I). In the fivestructures obtained, the analysis of solvent accessibilityand molecular cavities was carried out with programVOIDOO20 using a reduced atomic radius for polar atomsin accounting for possible hydrogen bonds.21

Structure factors and coordinates are accessible fromPDB with identification codes 1GGE, 1GG9, 1GGF, 1GGH,1GGJ, and 1GGK, respectively, for native HPII, His128Asn,His128Asn–H2O2, His128Ala, Asn201Ala, and Asn201Hisvariants.

RESULTSQuality of the Variant HPII Structures

Like the structure of native HPII,12,13 the crystal asym-metric units of the four HPII variants analyzed in thisstudy (His128Ala, His128Asn, Asn201Ala, and Asn201His)contain homotetramers with 3,012 amino acids. The N-terminal 26 residues of all subunits in the four variantsremain disordered, as in the native enzyme, and are notincluded in the refined structures. Crystallographic agree-ment R and Rfree factors and relevant refinement statisticsfor the structures of the four variants are similar to theones obtained for the native enzyme, particularly for theHis128Asn and the Asn201Ala variants determined atabout the same resolution as native HPII (Table I). Thequality and resolution of the diffraction data are slightly

STRUCTURES OF CATALASE HPII VARIANTS 271

Figure 1.

272 W. MELIK-ADAMYAN ET AL.

lower for the Asn201His variant structure, probably aresult of data collection at room temperature. The qualityand resolution of the His128Asn crystals soaked in H2O2

permitted accurate determination of differences in theprotein structure and in the solvent organization causedby the soaking.

Structural Differences of the HPII Variants WithRespect to the Native Enzyme

The absence of the imidazole group of His128, the distalside essential histidine of HPII, precludes the formation ofcompound I and results in all variants of His128 lackingcatalytic activity.14 The His128Ala variant is no exception,with no activity over background levels being detectableeven at very high protein concentrations (Table II). Thestructure of this variant reveals the expected methyl groupof Ala128 above ring IV of the heme b (note the propionateside-chain in Fig. 1B, as compared with the spirolactonering in the heme d of the native enzyme in Fig. 1A) in placeof the imidazole ring of histidine in native HPII. Despitethe increased accessibility to the iron atom in the variantresulting from the small size of the alanine residue, the

closest solvent molecule is ;4.0 Å from the heme iron, evenfarther away than in the native enzyme, indicative of thevery weak binding of water in the sixth coordinationposition. The presence of heme b, rather than heme d, andthe absence of the His392–Tyr415 covalent linkage areconsistent with the total absence of catalytic activity inthis variant.14 The propionate group from ring III of theunmodified heme b is hydrogen bonded with the nearbyGln419 residue, which also interacts with an adjacentwater molecule. An identical rearrangement had alreadybeen found in other HPII variants that contain heme b.10

The imidazole group of His392, not covalently linked toTyr415, is rotated with respect to the orientation seen inthe native enzyme (Fig. 1A,B). Other than the changes justdescribed and the extensively altered solvent organizationdescribed below, the His128Ala variant protein is essen-tially identical in structure to native HPII, particularly inthe heme distal side pocket.

The His128Asn variant also lacks all detectable cata-lytic activity, attributable to the absence of the essentialimidazole ring (Table II). Aside from the obvious presenceof the asparagine side-chain interacting with the main-

Fig. 1. Stereo views of the heme environment. A: Native HPII. B: His128Ala variant. C: Asn201His variant.For clarity, only the catalytically important residues His128, Ser167, and Asn201 on the heme distal side andHis392, Arg411, and Tyr415 on the heme proximal side are explicitly shown. Also displayed are the conservedresidues lining the channel, Val169, Asp181, Phe207, and Phe217. The ring of hydrophobic residues thatinclude Val169 define the narrowest point in the major channel. Heme d is evident only in native HPII (A), andthe covalent bond between the side-chains of His392 and Tyr415 is evident in native HPII (A) and theAsn201His variant (C). Changes in solvent organization are evident among the three structures. Watermolecules in the native enzyme, and their equivalent in the variants, are labeled numerically, W1–W8. Watermolecules in the variant structures with no correspondence in native HPII are labeled alphabetically, WA–WE.[Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.].


chain oxygen of residue 168, the protein structure of theHis128Asn variant is very similar to that of the His128Alavariant including the absence of the His392–Tyr415 cova-lent bond and the presence of heme b, with all the

concomitant readjustments described above for theHis128Ala structure.

The Asn201Ala variant exhibits about 10% of nativeHPII catalytic activity (Table II), indicating that residueAsn201 has a major, but not completely essential, role incatalysis. On the distal side of the heme, the most obviouschange in the protein structure of the variant is theexpected absence of the amide group of the asparagineside-chain. The heme is oxidized to heme d, consistent withthe previously reported biochemical characterization ofheme d in this variant,14 and the His392–Tyr415 bond isalso evident, making the structure in this region verysimilar to that of native HPII (Fig. 1A). The fact that bothheme oxidation and His–Tyr bond formation occur despitethe reduced catalytic activity, confirms that even a slowgeneration of compound I is sufficient to promote bothreactions.

TABLE I. Data Collection and Structural Refinement Statistics

Protein

Native H128A H128N N201H* N201A HP-H128N

A. Data collection statistics

Resolution range (Å)18/1.89 15/2.15 15/1.89 13/2.26 19/1.92 19.8/2.28

Unique reflections 214389 143588 196776 120733 193708 125885Completeness (%) 97.7 95.9 89.5 88.2 91.7 98.5^F&/^s(F)& 11.17 10.21 10.81 14.54 6.45 10.9Rsym (%) 8.9 10.5 8.7 8.9 13.9 10.7

In the last shellResolution range (Å) 1.94/1.89 2.21/2.15 1.94/1.89 2.32/2.36 1.97/1.92 2.34/2.28Unique reflections 16094 11135 16215 10095 15468 9488Completeness (%) 88.4 93.3 80.2 68.4 89.5 95.9^F&/^s(F)& 5.00 5.25 4.32 6.67 3.13 4.11Rsym (%) 26.9 26.4 27.5 35.5 49.0 36.5

B. Refinement statistics

Working set203615 136412 186868 114621 184003 119537

Free reflections 10744 7176 9907 6111 9705 6348Rcryst (%) 16.6 16.1 15.9 14.3 19.4 18.5Rfree (%) 20.2 20.6 18.8 20.8 23.5 25.4

In the last shellRcryst (%) 18.9 17.1 17.9 18.3 24.9 19.8Rfree (%) 23.2 22.8 21.9 27.1 28.8 29.6

No. of nonhydrogen atomsProtein 22984 22964 22984 22992 22984 22964Heme 176 172 172 172 176 172Water 3208 2828 3439 1723 2802 2036Hydrogen peroxide — — — — — 12

RMSD from idealityBond distances (Å) 0.008 0.008 0.007 0.010 0.008 0.009Angle distances (Å) 0.025 0.027 0.023 0.031 0.025 0.031Planarity (Å) 0.019 0.019 0.018 0.020 0.018 0.020Chiral volume (Å3) 0.114 0.110 0.103 0.121 0.103 0.117

Averaged B factors (Å2)Main-chain 10.6 11.0 10.4 19.2 8.6 24.2Side-chain 11.9 12.6 13.3 20.8 10.7 24.3Water 18.6 17.5 20.6 24.1 17.2 21.0

RMSD, root-mean-square deviation.

TABLE II. Catalatic Activity and PosttranslationalModifications

Variant Heme Tyr415–His392Activityunits/mg

Native d y 15,200His128Ala b n ,0.1His128Asn b n ,0.1His128Asn–H2O2 b n —Asn201Ala d y 1,300Asn201His b y 100


The Asn201His variant exhibits less than 1% of nativeHPII catalytic activity (Table II). This reduction can beexplained by a combination of the absence of the catalyticasparagine and by the reduction of the distal pocketvolume due to the presence of the bulkier imidazole groupin the variant. The Ne atoms of the two imidazole groupsfound in the heme distal side of the Asn201His variant(His128 and His201) are hydrogen bonded to each otherwith Ne from His128 acting as the hydrogen acceptor (Fig.1C). As a result, atom Nd from His201 is deprotonatedexplaining the absence of a hydrogen bond with watermolecule W3*. In native HPII, the corresponding watermolecule (W3 in Fig. 1A) stabilizes the appropriate orienta-tion of the Asn201 side-chain for it to act as a hydrogenacceptor. The Asn201His variant was characterized bio-chemically as containing only heme b,14 as confirmed inthe current structural analysis. Surprisingly, the struc-ture shows clear evidence for the presence of the His–Tyrbond (Fig. 1C), providing the first evidence that the twoself-catalyzed posttranslational modifications can be un-coupled in HPII.

Solvent Reorganization in the HPII Variants WithRespect to the Native Enzyme

In addition to the structural differences compared withnative HPII described above, the four active site variantsreveal striking reorganizations of the solvent both near thereplaced residues and in the main channel leading to theheme distal pocket (Figs. 1–3). Three of the variants,His128Asn, His128Ala, and Asn201Ala, all have a largernumber of water molecules in these regions than thenative enzyme, while the Asn201His variant, with abulkier residue introduced, contains fewer.

More significant than changes in the number of watersare changes in the distribution of the waters in thevariants as compared with native HPII. In most catalases,including HPII, there are two water molecules (indicatedas W1 and W2 in Fig. 1A) hydrogen bonded to each other,that bridge the essential histidine and asparagine resi-dues in the heme distal side pocket. W1, which acts as ahydrogen donor in its interaction with atom Ne fromHis128, is the solvent molecule closest to the heme ironatom (at a distance of ;2.5 Å) and in native HPII presentseither a low occupancy or a high mobility.13 The secondwater, W2, with full occupancy or low mobility, is bound toNd from Asn201. Three additional solvent molecules (W3,W4, and W5 in Fig. 1A), situated in close proximity to theheme, are also well conserved among catalases. Proceed-ing up the main channel away from the distal side hemepocket, the first well defined solvent molecule in the mainchannel (W6 in Fig. 1A) is approximately 7 Å from W2, andis hydrogen bonded to an aspartate residue (Asp181 inHPII) conserved in all catalase sequences. The absence ofany other well defined solvent molecules in the lower partof the main channel leaves a large gap between W2 andW6, which is apparently empty (Fig. 1A). From W6, achain of hydrogen bonded solvent molecules extends upthe channel to the molecular surface.

In the His128Ala variant (Figs. 1B, 2), replacement ofthe imidazole ring with the much smaller methyl side-chain of alanine increases the size of the active site cavity,but also removes a side-chain capable of forming hydrogenbonds. It is therefore somewhat surprising to find that thenumber of water molecules in this region increases substan-tially to seven from two in native HPII. The five watermolecules in the variant, without equivalent in the nativestructure, are indicated as WA to WE in Figure 1B. Theappearance of waters WA and WB in the narrowest andmost hydrophobic part of the channel provides continuityto the chain of solvent molecules extending from the hemedistal pocket to the molecular surface. The displacement ofW2*, with respect to the position of the correspondingwater W2 in the native enzyme, reduces the distancebetween W2* and W6, allowing water WB to interact withthe two.

The pattern of water molecules in the His128Asn vari-ant is very similar to that just described for the His128Alavariant, with the exception that water WE is missing,displaced by the larger asparagine side-chain at position128 (Fig. 3A). The presence of two asparagines (Asn128and Asn201) stabilizes the water molecules, filling thedistal side pocket and becoming part of a continuous chainof waters extending to the molecular surface (Fig. 3A).Waters WA and WB in the His128Asn variant are inalmost identical locations to the corresponding watersfound in the His128Ala variant. Again the reduced dis-tance between W2* and W6 is bridged by the single waterWB.

The replacement of asparagine by histidine in theAsn201His variant creates yet another striking organiza-tion of solvent molecules (Fig. 1C). In the heme distal sidepocket, there is only one defined water molecule, at aboutthe same position as W1 in the native enzyme, likely aresult of the reduction in pocket volume arising from thelarger size of the imidazole side-chain. Even more surpris-ing is the altered solvent structure around the Asp181residue, remote from His201, which clearly demonstratesthe subtle interdependencies among the solvent bindingsites.

The Asn201Ala variant has a poorly defined solventstructure in the vicinity of the heme. There are severalwater molecules with large mobility or partial occupan-cies, probably due to the increased cavity size and absenceof a side-chain capable of forming hydrogen bonds. Signifi-cantly, W3* in this variant is in an almost identicalposition to W3 in the native enzyme, despite the lack ofinteractions with the replaced residue Ala201. The stabil-ity of W3 provides further support for its critical role indefining the orientation of the side-chain of Asn201 in thenative enzyme (see above). Despite the increased cavityvolume, the absence of a stable W2* seems to prevent awater from occupying position WB. These data confirmthat minor changes in the heme distal side pocket canalternatively stabilize or destroy the continuity of solventmolecules in the hydrophobic part of the channel.


Fig. 2. Stereo views showing the opening of the major channel into the heme distal side pocket in theHis128Ala variant structure. A: Averaged omit (Fo 2 Fc) electron-density map shown with a chicken wirerepresentation. Density corresponding to the chain of omitted water molecules is clearly defined filling theheme distal side pocket and the major channel. B: The accessible surface, calculated with the programVOIDOO, emphasizes the continuity of the channel and of the chain of solvent molecules found inside. Redspheres, water molecules; dashed lines, hydrogen bonds among water molecules in the channel; dotted lines,iron coordination.

His128Asn HPII Variant in Complex With H2O2

It has not yet been possible to trap hydrogen peroxideinside the structure of active catalases because of the rapidreaction and evolution of O2 causing serious crystal dam-age upon soaking in hydrogen peroxide solutions. How-ever, using inactive catalase variants could circumventthis problem. Diffraction data from flash cooled crystals ofthe inactive His128Asn variant, previously soaked for afew seconds in a 2M hydrogen peroxide solution, wereobtained at 2.3-Å resolution. Preliminary analysis of thesecrystals, using (Fo

complexed 2 Founcomplexed) difference Fou-

rier maps, indicated the presence of several significantstructural changes with respect to the structure of theuncomplexed variant (Figs. 3, 4). Therefore, the structureof the complex was refined, following the same protocols asthose described above, and using as a starting model theuncomplexed His128Asn variant structure with all thesolvent molecules in the distal pocket and in the mainchannel removed (Table I). A diversity of (2Fo 2 Fc) and(Fo 2 Fc) maps, both unaveraged or fourfold averagedusing noncrystallographic symmetry, were obtained. Tocompute these maps, the solvent molecules to be analyzedwere either omitted or included as water or hydrogenperoxide. All these data provided a consistent support forthe presence of three hydrogen peroxide molecules invirtually the same locations in all four crystallographicallyindependent subunits (Table III). Two hydrogen peroxidemolecules, H1 and H2, are located immediately above theheme interacting with Asn128 and Asn201, and the thirdperoxide, H3, is located in the main channel interactingwith Ser234 and Glu539 (Fig. 3B). The B values of these 12peroxide molecules are all small in comparison to the meansolvent B factor (Table III), and they provide a cleanexplanation for the electron density observed both in theheme distal pocket and further up the main channel. Fromthe location of H3, there is a direct connection between themain channels of the two subunits related by the molecu-lar R-axis.22

Unexpectedly, the greatest structural changes in theprotein were found remote from these three peroxidemolecules on the heme proximal side, clustered around thecentral cavity in the tetramer (Fig. 4). These changesinvolve, in particular, residues Arg111, Phe413, Thr416,and Asp417, in the vicinity of Tyr415, the iron proximalligand. Even more unexpected, was the observation that

the changes occur in only two subunits (related by theQ-axis) causing a deviation from the tetrameric symmetryof the native enzyme; in effect; the complex becomes adimer of structurally nonidentical dimers. Residues Ile129,Met448, Cys438, and Pro439, in the vicinity of the centralcavity, also change position, but without affecting symme-try. Solvent molecules in the vicinity of all the displacedresidues experience important rearrangements, and afourth hydrogen peroxide molecule might be situated closeto Ile129, although it is not included in the present model.The presence of the fourth peroxide suggests a possiblepath leading from the proximal side of the heme groups tothe center of the tetramer, a path required by the proposedmechanism for heme oxidation and His–Tyr bond forma-tion.11,21

H1, the H2O2 situated closest to the heme, interactswith Od from Asn128 and with H2, the second H2O2, and issituated between the positions corresponding to the pair ofwater molecules WC*–W1* in the structure of the uncom-plexed variant (Fig. 3A). H2, the best-defined H2O2 (TableIII) interacts with H1, as already indicated, with the Nd ofAsn128, and with both Od and Nd from Asn201. H2 issituated between the positions corresponding to the pair ofwater molecules WD–W2 in the uncomplexed variant.Distances between H1 or H2 and W6 are too large to bebridged by a single solvent molecule, and no solvent, eitherhydrogen peroxide or water, is evident in the lower hydro-phobic part of the major channel. Therefore, a high solventoccupancy in this portion of the channel seems to requirethe additional stabilization provided by the hydrogenbonded matrix of a chain of solvent molecules as in theuncomplexed His128Asn variant.

DISCUSSIONMechanistic Uncoupling of Heme Oxidation fromHis–Tyr Bond Formation

Previous work had suggested a close link between hemeoxidation and His–Tyr bond formation in HPII, and aconcerted mechanism was proposed.11 Therefore, it was amost unexpected observation to find the His–Tyr bond andunoxidized heme b in the Asn201His variant. Alternativemechanisms treating the two modifications as separatereactions had been considered11 and now appears to be themost likely option in view of this result. The starting pointfor both modifications continues to be the formation ofcompound I because its absence, as in the His128 variants,prevents both reactions. The Asn201His variant exhibitsonly a very low level of activity, which presumably means alow level of compound I formation. The fact that this lowlevel of compound I is sufficient to promote the formationof the His–Tyr bond, but not heme oxidation, suggests thatthe energy barrier for His–Tyr bond formation is lowerthan for heme oxidation. One implication of this conclusionis that the cis-stereospecific heme oxidation should not beobserved in the absence of His–Tyr bond formation; todate, this is the case.21 Treatment of the Asn201Hisvariant with either ascorbate or glucose–glucose oxidase,both of which generate low levels of hydrogen peroxide,causes a spectral change from a heme b-type spectrum to a

TABLE III. Temperature Factors (Å2) of the HydrogenPeroxide Molecules

H2O2 Atom

Subunit

1 2 3 4

H1 O1 29.0 25.5 35.7 22.0O2 29.1 28.6 34.8 21.7

H2 O1 16.1 17.9 20.5 18.3O2 13.8 19.0 21.4 18.5

H3 O1 22.6 31.3 18.8 23.4O2 25.1 32.3 18.7 21.7


heme d-type spectrum.14 Despite this evidence for heme dformation, only heme b could be isolated from the variant,never heme d. Therefore, the reaction pathway to hemeoxidation can be initiated and can produce a molecularspecies that is spectrally similar to heme d, but whichcannot be fully converted to heme d and reverts to heme bon extraction in acetone-HCl.

Solvent Flow in Catalase

Small ligands of metalloproteins, such as NO, CO, O2, orH2, have important functions in signal transduction, respi-ration, and catalysis. The question of whether ligandaccess to active centers requires specific channels anddocking sites or proceeds by random diffusion through theprotein matrix has recently been addressed with theconclusion that there are a limited number of pathways forligand migration.23 Similar questions are also relevant tocatalase, where the rapid turnover rate requires an effi-cient mechanism for H2O2 to gain access to, and forproducts to be exhausted from, the deeply buried activesites. In catalase, the existence of alternative channelssuggests a possible directional flow model in which onechannel is used for substrate intake and a second (orpossibly more than one) is used for product exhaust. Themajor channel, which enters the heme distal side pocketperpendicular to the heme surface, is highly conservedamong all catalases and has long been considered theprime candidate for the inlet channel, although supportingevidence is limited. In small subunit enzymes, like BLC,this channel is ;35 Å in length, but in large subunitenzymes, like HPII, the normal entrance is partiallyoccluded by the C-terminal extension and may be 20 Ålonger.11 In both types of catalase, the inner part of themajor channel narrows into a constricted region lined withhydrophobic residues that opens into the heme distal sidepocket (Figs. 1–3). The constricted hydrophobic portion ofthe channel, spanning ;7 Å, makes access difficult formolecules larger than water and H2O2, presumably helpsdetermine the substrate specificity of the enzyme, andprevents access of potential inhibitors. Recently this por-tion of the channel has been referred to as a “molecularruler” as part of a hypothesis that the channel’s dimen-sions are optimized for hydrogen peroxide,3 but not water,explaining the absence of well-defined water in this sectionof the channel.

The HPII variants described in this article demonstratethat any perturbation in the active site of HPII whichstabilizes the inclusion of more waters in the cavity canresult in solvent occupancy of the hydrophobic portion ofthe major channel and in the formation of a continuouschain of water, .30 Å in length, extending from themolecular surface to the active site. In fact, only one or twoadditional hydrogen bonds are sufficient to stabilize sol-vent (WA or WB) in the hydrophobic section of the channel.Consequently, the energy barrier to solvent residency inthe hydrophobic portion of the major channel is, at most, afew kilojoules per mole (kJ/mol). The larger size of hydro-gen peroxide and its greater hydrophilic character com-pared with water might eliminate this barrier in native

catalases, permitting the formation of a hydrogen-bondedchain extending through the hydrophobic segment of thechannel. The stereo-specific requirements underlying theformation of a continuous chain of solvent molecules canhelp explain the substrate specificity of the enzyme, aswell as the apparent fragility of the structural feature.

The unidirectional movement of a chain of substratemolecules, which would facilitate the selective access ofH2O2 molecules into the active site as the catalytic reac-tion progresses, requires a separate channel for the ex-haust of reaction products. The existence of separate inletand exhaust channels leading to and from the active siteallows a “flow” of molecules through the catalase withH2O2 entering by one channel and H2O and O2 exiting bydifferent ones. The drive for such a flow could be providedby the highly exothermic catalytic reaction (Go9 of 2209kJ/mol).24

Substrate flowing through catalase is conceptually analo-gous to solute flowing through membrane transport pro-teins, and the similarity is strengthened when the struc-tures of catalases and some porins are compared. Bothcatalases and porins have a b barrel core surrounding afunnel shaped channel leading to a narrow hydrophobicconstriction, which provides specificity to the substratebeing transported.25 The hydrophobic region of the porinchannel has been implicated as a facilitator, or “greaseslide,” for the flow of hydrophilic metabolites,25 and thehydrophobic constriction of catalase may be ascribed asimilar role in favoring H2O2 movement. The channellengths are similar in both types of proteins with a porinspanning a 50-Å membrane compared with the approxi-mately 60 Å combined length of the proposed inlet and exitchannels in a catalase. Porins provide an efficient flow ofspecific substrates across the membrane, and catalasesprovide an efficient flow of H2O2 to the active site and of O2

and H2O to the outside. The importance of channel archi-tecture leading to the active center is illustrated by the2.5-fold increase in turnover rate in HPII resulting fromthe single change of Arg260 to Ala located 20 Å from theheme in the minor or lateral channel, a possible candidatefor the exhaust channel.26

Broken Molecular Symmetry

The observation that molecular symmetry of the variantbreaks during soaking with H2O2 is a most intriguing

Fig. 3. Stereo views of the heme distal side of the His128Asn variantwhen free (A) or complexed with hydrogen peroxide (B). The solventorganization in the structure of the free variant is closely related to the onefound in His128Ala (Fig. 2B). Hydrogen peroxide molecules found in thecomplex are labeled H1, H2, and H3 (B). Water labeling is as describedfor Fig. 1. The averaged omit (Fo 2 Fc) electron-density maps are bothpresented at 2.3-Å resolution to facilitate the comparison. Differences inthe solvent organization in the two structures are evident. The electrondensities corresponding to the three putative H2O2 molecules present anelongated shape with volumes and electron-density values that cannot beexplained by simple combinations of water molecules (see text). Each ofthe three putative H2O2 molecules (blue) is situated between positionsthat correspond to the positions of two water molecules in the uncom-plexed variant. [Color figure can be viewed in the online issue, which isavailable at www.interscience.wiley.com.]


Figure 3.


Fig. 4. Views down the molecular P (A), Q (B), and R (C) axes comparing the structures of the His128Asn variant free and complexed with H2O2.Residues exhibiting the largest movement as a result of peroxide binding are displayed with a CPK representation. The heme residues are colored red.The subunits are color-coded: A, yellow; B, red; C, green; and D, cyan. Letters correspond to the PDB designations. The striking clustering of differencesin subunits on one side of the cavity situated in the center of the molecular tetramer suggests an active role for the heme proximal side during catalysis.The departure from the original 222 molecular symmetry of the uncomplexed variant is also evident.

result that requires further investigation. Inter-subunitcontacts between His449 residues along the molecular axisin native HPII,11 already require small departures fromperfect molecular symmetry and could be the trigger of anasymmetric behavior during catalysis. Deviations fromperfect symmetry have also been reported for other cata-lases, particularly for HEC, where NADP(H) appearsbound to only two of the subunits and where, in addition,compound I formation was shown to take place only in thetwo subunits lacking NADP(H). The apparent asymmetricstructural features are consistent with classical kineticstudies that suggest only half of the heme groups areactive simultaneously.27 Hence, the departure from sym-metry arising from peroxide binding is another indicationthat molecular symmetry can be easily broken duringenzyme catalysis, which may allow for an unusual type ofcooperativity among subunits. This, of course, leads to theunderlying question as to how H2O2 binding and catalyticactivity in two subunits interfere with or prevent activityin the other two subunits—a question that remains unan-swered.

ACKNOWLEDGMENTS

Work in Barcelona was funded by grant BIO099-0865from DGICYT (Spain) and by the European Union throughthe HCMP to Large Installations Project (contract CHGE-CT93-0040). Work in Winnipeg was supported by grantRGP9600 from the Natural Sciences and EngineeringResearch Council of Canada (NSERC). A NATO Collabora-tive Research Grant (SA-5-2-05) (to P.C.L. and I.F.) alsosupported the work. Fellowship SAP 1998-0087 from MEC(Spain) was awarded to W.M.A.

REFERENCES

1. Murthy MRN, Reid TJ, Sicignano A, Tanaka N, Rossmann MG.Structure of beef liver catalase. J Mol Biol 1981;152:465–499.

2. Fita I, Silva AM, Murthy MRN, Rossmann MG. The refinedstructure of beef liver catalase at 2.5 Å resolution. Acta Crystal-logr B 1986;42:497–515.

3. Putnam CD, Arval AS, Bourne Y, Tainer JA. Active and inhibitedhuman catalase structures: ligand and NADPH binding andcatalytic mechanism. J Mol Biol 2000;296:295–309.

4. Ko TP, Safo MK, Musayev FN, Di Salvo ML, Wang C, Wu, SH,Abraham DJ. Structure of human erythrocyte catalase. ActaCrystallogr D Biol Crystallogr 2000;56:241–245.

5. Vainshtein BK, Melik-Adamyan WR, Barynin VV, Vagin AA,Grebenko AI, Borisov VV, Bartels KS, Fita I, Rossmann MG.Three-dimensional structure of catalase from Penicillium vitale at2.0-Å resolution. J Mol Biol 1986;188:49–61

6. Melik-Adamyan WR, Barynin VV, Vagin AA, Borisov VV, Vainsh-tein BK, Fita I, Murthy MRN, Rossmann MG. Comparison of beefliver and Penicillium vitale catalases. J Mol Biol 1986;188:63–72.

7. Mate MJ, Zamocky M, Nykyri LM, Herzog C, Alzari PM, Betzel C,Koller F, Fita, I. Structure of catalase-A from Saccharomycescerevisiae. J Mol Biol 1999;286:135–149.

8. Gouet P, Jouve HM, Dideberg O. Crystal structure of Proteusmirabilis PR catalase with and without bound NADPH. J Mol Biol1995;249:933–954.

9. Murshudov GN, Melik-Adamyan WR, Grebenko AI, Barynin VV,Vagin AA, Vainshtein BK, Dauter Z, Wilson KS. Three-dimen-sional structure of catalase from Micrococcus lysodeikticus at 1.5 Åresolution. FEBS Lett 1992;312:127–131.

10. Bravo J, Verdaguer N, Tormo J, Betzel C, Switala J, Loewen PC,Fita I. Crystal structure of catalase HPII from Escherichia coli.Structure 1995;3:491–502.

11. Bravo J, Mate MJ, Schneider T, Switala J, Wilson K, Fita I,Loewen PC. Structure of catalase HPII from Escherichia coli at1.9Å resolution. Proteins 1999;34:155–166.

12. Murshudov GN, Grebenko AI, Barynin V, Dauter Z, Wilson KS,Vainshtein BK, Melik-Adamyan W, Bravo J, Ferran JM, FerrerJC, Switala J, Loewen PC, Fita I. Structure of the heme d ofPenicillium vitale and Escherichia coli catalases. J Biol Chem1996;271:8863–8868.

13. Bravo J, Fita I, Ferrer JC, Ens W, Hillar A, Switala J, Loewen PC.Identification of a novel bond between a histidine and the essen-tial tyrosine in catalase HPII of Escherichia coli. Protein Sci1997;6:1016–1023.

14. Loewen PC, Switala J, von Ossowski I, Hillar A, Christie A,Tattrie B, Nicholls P. Catalase HPII of Escherichia coli catalyzesthe conversion of protoheme to cis-heme d. Biochemistry 1993;32:10159–10164.

15. Otwinowski Z, Minor W. Processing of X-ray diffraction datacollected in oscillation mode. Methods Enzymol 1996;276:307–326.

16. Brunger AT. XPLOR Manual. Version 3. New Haven, CT: YaleUniversity; 1992.

17. Murshudov GN, Vagin AA, Dodson EJ. Refinement of macromo-lecular structures by the maximum-likelihood method. Acta Crys-tallogr D Biol Crystallogr 1997;53:240–255.

18. Perrakis A, Morris RJH, Lamzin VS. Automated protein modelbuilding combined with iterative structure refinement. NatureStruct Biol 1999;6:458–463.

19. Jones TA, Zou JY, Cowan SW, Kjeldgaard M. Improved methodsfor building protein models in electron density maps. Acta Crystal-logr A 1991;47:110–119.

20. Kleywegt GJ, Jones TA. Detection, delineation, measurement anddisplay of cavities in macromolecule structures. Acta CrystallogrD Biol Crystallogr 1994;50:178–185.

21. Mate MJ, Sevinc MS, Hu B, Bujons J, Bravo J, Switala J, Ens W,Loewen PC, Fita I. Mutants that later the covalent structure ofcatalase hydroperoxidase II from Escherichia coli. J Biol Chem1999;274:27717–27725.

22. Nicholls P, Fita I, Loewen PC. Enzymology and structure ofcatalases. Adv Inorg Chem 2001;51:51–106..

23. Chu K, Vojtchovsky J, McMahon BH, Sweet RM, Berendzen J,Schlichting I. Structure of a ligand-binding intermediate in wild-type carbonmonoxy myoglobin. Nature 2000;403:921–923.

24. Nicholls P, Schonbaum GR. Catalases. The enzymes. Vol 8. 2nded. New York: Academic Press; 1963. p 147–217.

25. Forst D, Welte W, Wacker T, Diederichs K. Structure of thesucrose-specific porin ScrY from Salmonella typhimurium and itscomplex with sucrose. Nature Struct Biol 1998;5:37–45.

26. Sevinc MS, Mate MJ, Switala J, Fita I, Loewen PC. Role of thelateral channel in catalase HPII of Escherichia coli. Protein Sci1999;8:490–498.

27. Chance B, Oshino N. Kinetics and mechanisms of catalase inperoxisomes of the mitochondrial fraction. Biochem J 1971;122:225–233.


Substrate flow in catalases deduced from the crystal structures of ...

Documents

Transcript of Substrate flow in catalases deduced from the crystal structures of ...